Short-lived ice speed-up and plume water flow captured by a VTOL UAV give insights into subglacial hydrological system of Bowdoin Glacier

Abstract

The subglacial hydrology of tidewater glaciers is a key but poorly understood component of the complex ice-ocean system, which aects sea level rise. As it is extremely dicult to access the interior of a glacier, our knowledge relies mostly on the observation of input variables such as air temperature, and output variables such as the ice ow velocities reecting the englacial water pressure, and the dynamics of plumes reecting the discharge of meltwater into the ocean. In this study we use a cost-eective Vertical Take-O and Landing (VTOL) Unmanned Aerial Vehicle (UAV) to monitor the daily movements of Bowdoin Glacier, northwest Greenland, and the dynamics of its main plume. Using Structure-from-Motion photogrammetry and feature-tracking techniques, we obtained 22 high-resolution ortho-images and 19 velocity elds at the calving front for 12 days in July 2016. Our results show a two-day-long speed-up event (up to 170%) caused by an increase in buoyant subglacial forces with a strong spatial variability revealing that enhanced acceleration is an indication of shallow bedrock. Further, we used the Particle Image Velocimetry (PIV) method to analyze water ow from successive UAV images taken while ying over the main plume of the glacier. We found that PIV successfully captures the area of radially diverging ow of the plume, and provides information on spatial and time variability as no other remote sensing technique can. Most interestingly, the active part of the plume features pulsating water jets at the time scale of seconds, and is 1 to 5 times smaller than its visual footprint dened by the iceberg-free area. Combined with an ice ow model or a non-steady plume model, our approach has the potential to generate a novel set of input data to gather information about the depth of the bedrock, the discharge of meltwater, or the subglacial melting rate of tidewater glaciers.

High accuracy UAV photogrammetry of ice sheet dynamics with no ground control

“…The case studies of a calving front and ice sheet interior provided in this study show two different applications of the GNSS-AT method: one in a calving front environment where UAVs have previously been used, and one in an ice sheet interior, where UAVs have not to date been used to assess ice dynamics. The first case study highlights that existing observations 10 of, for instance, calving events (Ryan et al, 2015;Jouvet et al, 2017Jouvet et al, , 2018 can be successfully replicated using GNSS-AT methods (Fig. 5).…”

“…However, it has been shown that the error of a GNSS-AT-derived product can be further constrained by the reintroduction of at least one GCP, with a final vertical accuracy only slightly worse than traditional GCP networks (Benassi et al, 2017). Whilst constructing a comprehensive network of GCPs might be difficult on glacial terrain, the introduction of one 15 GCP, either in the form of an existing continuous GPS station, or a single target measured on a per-flight basis or interpolated linearly as per Jouvet et al (2017Jouvet et al ( , 2018, is far more achievable than a large, dense network of GCPs. The method as described here also lacks the incorporation of directional data in the camera coordinate positions, and hence is referred to as GNSS-AT rather than full DSO (section 2.3).…”

“…Further work on UAV dynamics studies of calving fronts was developed by Jouvet et al (2017Jouvet et al ( , 2018 for Bowdoin Glacier 20 in North Greenland. They report an improved horizontal error of 10-20 cm, a value that improves on Ryan et al (2015), and approximately double that in this study.…”

“…), which improves the quality of photogrammetric reconstruction whilst limiting the 25 total area able to be assessed in a single flight (Bowdoin is ∼3 km across, whilst Store is ∼5 km). Hence, Jouvet et al (2017Jouvet et al ( , 2018 show that it is possible to work with moving one-ice GCPs to provide viable products. However, the logistical effort is still considerable, and as a result GCP density is sparse, with large distances (up to 2 km) between GCPs, which likely leads to significant errors at points far from GCP location (Tonkin and Midgley, 2016;Gindraux et al, 2017).…”

High-Endurance UAV for Monitoring Calving Glaciers: Application to the Inglefield Bredning and Eqip Sermia, Greenland

“…• tidal driven variability of the ice flow of ocean-terminating glaciers (Sugiyama et al, 2015) or the tidal flexure of ice shelves, which be can used to locate the grounding line (Le Meur et al, 2014), • short-lived speed-up events caused by a abrupt change in the subglacial hydrological system (e.g., Jouvet et al, 2018), • fracturing processes, such as small-scale calving (Walter et al, 2019) or the rifting of a tabular icebergs (Joughin and MacAyeal, 2005;Bassis and Jacobs, 2013;Medrzycka et al, 2016).…”